[Corpora-List] unsupervised with semi-supervised

[Taras Zagibalov]

Thank you Ted for your input.
I will summarise all the replies I will (hopefully) get.
As for the first question, I meant that all labelling was done manually. 
And the question is if the systems are unsupervised or semi-supervised, 
since some definitions state that semi-supervised systems make use of 
both labelled and unlabelled data. The confusing part is that although 
the amount of labelled data is so different, it doesn't seem to be relevant.
As for question 3. I do realise that rule-based systems are supposed to 
be in another paradigm. But this also confuses me, since rules can be 
used to produce a set of labelled examples and, as you noted, can be 
regarded as a form of supervision.

Regards,
Taras

Ted Pedersen wrote:
> If you did get some responses on your original question to you
> directly (and not to
> the list) it would be quite helpful if you posted as a summary of
> those to the list.
>
> In your question 1) did you mean to ask if the systems were still
> supervised? All
> those labeled examples suggest that they wouldn't be unsupervised
> (although I may
> have missed something in your query). Were the examples labeled manually or
> via some automatic means?
>
> On 3, I'm not sure why rule-based is being associated with supervised
> - do you mean
> that the creation of rules is a form of supervision? That seems
> reasonable, although
> often rule-based and supervised systems are treated as separate categories.
>
> Cordially,
> Ted

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora